Boss Dimension C chorus

[A.S.P.]

thank you, Sean, for this explanation here!

referring to another ("harm*nizer"-thread) , I wouldn`t have thought that the "addition" fact would matter much in low frequency-alterations as in a Vibrato, but would rather manifestate itself at drastic frequency-variations,
but it makes sense, the way you described it for simultaneously up & down shift
even at low amounts.

would you say it`d make more sense to "univibe" the phaseshift-stages (half a "Dome"-filter...) with staggered coefficients for evenly spreaded phaseshift throughout the bandwidth of interest, to make such a phaseshift-Chorus more musical?

[MartyMart]

This all reminds me of my favourite "preset" on an old Yamaha SPX90
"Pitch change C"
This was a detune in both directions of about 8/10 cents, it wasn't "moving"
but it provided a "lush" sound, great for instruments and backing vocals too !

..... I dont have one anymore .....

But my Boss DC-2 and PS-3 pitch shifter, get very close

Marty.

[Dave_B]

Is anyone aware of another pic of the circuit board?  This one is good: http://www.modezero.com/boss_dimension_c.htm and gives some clues on parts placement.  Another one from a different angle could help a great deal if someone was trying to do a layout.

[nelson]

Is anyone aware of another pic of the circuit board?  This one is good: http://www.modezero.com/boss_dimension_c.htm and gives some clues on parts placement.  Another one from a different angle could help a great deal if someone was trying to do a layout.

Really, that doesnt look all that bad.

Tight layout yeah, but not outwith the realm of DIY at all.

Could certainly fit it into a 1590BB.

[Steben]

Yep I guess 1590BB should work.

note the vertical mounted opamps. That helps too. I guess its also easier to design a PCB with in-line chips?

[SeanCostello]

thank you, Sean, for this explanation here!

would you say it`d make more sense to "univibe" the phaseshift-stages (half a "Dome"-filter...) with staggered coefficients for evenly spreaded phaseshift throughout the bandwidth of interest, to make such a phaseshift-Chorus more musical?

I don't know, to tell you the truth. Most of the theoretical explanations concerning phase shifting that I have read assume that all of the stages have the same coefficient, which is the closest approximation to a delay line (which is what you want for pitch shifting). I have not experimented with first order allpass stages with different coefficients in the digital realm, as it is a pain to program that digitally. Computing a phase shift coefficient using the bilinear z-transform usually requires a divide, which is expensive, but if the same coefficient is used for all of the stages, you can get away with it (especially if you only compute the coefficient once per block of samples, and use linear interpolation to smooth out the changes). If you need a separate coefficient for each stage, the computation gets pricey, and you probably want to look at a table lookup solution for the coefficient computation.

I haven't done any experimentation with this in the analog realm, either, except for playing around with my Voodoo Vibe (shameful secret: I suck at soldering, and haven't had much time for analog construction in the last few years). The Voodoo Vibe allows you to add an offset to the modulation, which enables you to "fine tune" whether the sweep takes place in an area where the harmonic relationships of the note are preserved or whether it sounds out of tune. Even with the Voodoo Vibe, you can hear the detuning between harmonics with low guitar notes.

Has anyone here done any work on what the spacing of the allpass sections in the Univibe actually does? Is it related to Dome filter theory?

One interesting note: This past spring, Jonathan Abel (of Universal Audio) and Julius Smith (of CCRMA) gave a talk at an Acoustical Society of America convention about using cascaded 2nd order allpass sections to get a specific group delay. This will probably first be useful in spring reverb emulation, but it might be useful for pitch modulation if the allpass sections are modulated, as changing the group delay will result in phase shifts/frequency shifts.

Sean Costello

P.S. By the way, frequency shifitng in opposite directions just produces tremolo. A single sideband shifter in one direction plus a single sideband shifter in another direction equals ring modulation.

[A.S.P.]

P.S. By the way, frequency shifitng in opposite directions just produces tremolo

if you have a stereo-set: up-left/down-right... ?

[SeanCostello]

P.S. By the way, frequency shifitng in opposite directions just produces tremolo

if you have a stereo-set: up-left/down-right... ?

Sorry, I meant when the signals were summed together electronically. Stereo is totally different, and totally cool.  For low shift amounts, sending upshift to one speaker and downshift to another results in the sound rotating around your head and through your skull. For high shift amounts, it ends up sounding like ring modulation, but with nice spatial characteristics.

Obviously, the cheapest way of doing this is with a single frequency shifter with both upshift and downshift outputs. The difference between the two outputs is one subtracts while the other adds, so you are talking about 2 opamp stages rather than one, or 2 ALU operations rather than 1 in DSP, so it is worth sticking in there.

The stereo tricks listed above work for barberpole phasers based around frequency shifters as well.

Sean Costello

[SeanCostello]

Hanert explained it here:
http://www.pat2pdf.org/patents/pat2905040.pdf
http://www.pat2pdf.org/patents/pat2382413.pdf
well - artificial delay lines - but to the same effect:
pitch-shift by phase-shift.

Nice patents! One thing to note is that, even though the scanning vibrato mentioned in both patents relies upon the phase shift qualities of low pass filters, the actual circuit is not very close to the standard phase shifter, but rather to a linearly interpolated delay line. Each L/C section is said to have "approximately" linear group delay (the relationship of phase shift to frequency), so crossfading between the filter sections will result in a sound much closer to delay line modulation than a phase shifter. So the harmonic relationship between partials in the input signal will pretty much be maintained. My guess is that the same sort of modulation tricks that work with BBDs will work with the scanning vibratos.

Of course, the thing is dissimilar to a linearly interpolated delay line, in that each "sample" of the delay is in fact a filter with a nonlinear phase response. It is "approximately" linear, but the approximation will result in some interesting deviations. A digital implementation would be interesting, as each section could have more delay than a single sample. The lowpass filters could probably be replaced with allpass sections for an interesting effect (I don't know if allpass filters had been developed at the time of the patents).

Sean Costello

[SeanCostello]

Reading the first patent more carefully...wow! Triangle wave antiphase chorus, using moving TAPE HEADS, back in the 1940's! I wonder if such a device was ever commercially released. And the scanner vibrato that implements the antiphase chorus is super cool as well.

Plus, this patent has the introduction of the tri-phase chorus used in string ensembles in the 1970's, although that seems to be discussed with having 3 separate banks of oscillators. The Korg Lambda used this: 3 separate divide-down chains, with a tri-phase LFO providing an ensemble effect. However, this works largely due to having a slight amount of detuning between the oscillators, in addition to the modulation from the tri-phase LFO. If you just have a fast tri-phase LFO for 3 delay lines, it sounds pretty bad - more like a really rapid beating effect than a vibrato. You need to have the slower tri-phase LFO added into the modulation signal to decorrelate the delay lines, so that the faster LFO is working on detuned sources. Crappy explanation, but a little bit of experimentation will reveal to the ears what I am explaining poorly in words.

Sean Costello

[Processaurus]

This all reminds me of my favourite "preset" on an old Yamaha SPX90
"Pitch change C"
This was a detune in both directions of about 8/10 cents, it wasn't "moving"
but it provided a "lush" sound, great for instruments and backing vocals too !

A local recording engineer told me an interesting trick thats like that for multi-tracking the same part, after tracking once, you change the recording pitch up on your recorder a cent or so, and record your part again, and then pitch it down the same amount, and record again.The resulting mix is still in tune, just thicker.  You can do that as many times as you need, as long as the tracks are equally out of tune up and down from the original.

My favorite chorus sound is sadly also from digital pitch shifters with a detune function, too.  Because it doesn't wriggle around, I guess.

One thing I've wondered, that is hard to measure, is in a BBD chorus/vibrato circuit if pitch is related in a linear way to the LFO driving the high frequency clock.  For instance if a perfect linear triangle wave is fed to the clock, will the vibrato pitch stay the same amount sharp or flat when the wave is ramping up and down, and change instantly when it peaks?   

[Dave_B]

A local recording engineer told me an interesting trick thats like that for multi-tracking the same part, after tracking once, you change the recording pitch up on your recorder a cent or so, and record your part again, and then pitch it down the same amount, and record again.The resulting mix is still in tune, just thicker.  You can do that as many times as you need, as long as the tracks are equally out of tune up and down from the original.

I used to do that routinely with my Fostex Model 80 tape recorder, though to a lesser degree.  Record the guitar, rotate the "pitch control" knob slightly, then double the part on another track.  Then I panned the two tracks to about 10 o'clock and 2 o'clock.  The great thing is that you could hear the results immediately and adjust the depth of the effect as you messed around with the second track. 

I remember reading that Jake E. Lee recorded his parts four times to thicken the sound.  That seems a bit much, but who am I to say.   

One thing I've wondered, that is hard to measure, is in a BBD chorus/vibrato circuit if pitch is related in a linear way to the LFO driving the high frequency clock.  For instance if a perfect linear triangle wave is fed to the clock, will the vibrato pitch stay the same amount sharp or flat when the wave is ramping up and down, and change instantly when it peaks?

Are you referring to triangular vs. hypertriangular?  Or are you thinking the DimC uses a trianglular LFO (I'm pretty sure it's square).

[A.S.P.]

Hanert explained it here:
http://www.pat2pdf.org/patents/pat2905040.pdf
http://www.pat2pdf.org/patents/pat2382413.pdf
well - artificial delay lines - but to the same effect:
pitch-shift by phase-shift.

Nice patents!
Sean Costello

yes,that should have read frequency-shift rather than phase...

and: yes, the art was aware of Allpass-filters at the time (Harold Bode - don`t know, whether he was related to Harald B.).

[Jaicen_solo]

Ok,
Now I don't really understand the circuit diagrams of BBD based effects very well, so I'm hoping someone can help me out.
I've got myself a nice Yamaha DX27 for christmas. Unfortunately it has only Mono outputs so i'm looking for a good chorus to thicken up the pads I want it for. In this case, a Dimension would be just what i'm looking for.
What I need to know is if it is possible to modify FP's CE-2 layout to add a second BBD driven by an inverted clock. I have a couple of MN3007's, so I'd rather use them than the SAD1024 used in the Small Clone etc.
I remember the difference the ensemble effect used to make on the old Juno 106 we had at uni, it was literally like night and day! Everything just seemed so cold and one dimensional without it!
So, anybody got any pearls of wisdom??

[Steben]

Are you referring to triangular vs. hypertriangular?  Or are you thinking the DimC uses a trianglular LFO (I'm pretty sure it's square).

Square  I don't think so. Have to check again.

[Steben]

Ok,
Now I don't really understand the circuit diagrams of BBD based effects very well, so I'm hoping someone can help me out.
I've got myself a nice Yamaha DX27 for christmas. Unfortunately it has only Mono outputs so i'm looking for a good chorus to thicken up the pads I want it for. In this case, a Dimension would be just what i'm looking for.
What I need to know is if it is possible to modify FP's CE-2 layout to add a second BBD driven by an inverted clock. I have a couple of MN3007's, so I'd rather use them than the SAD1024 used in the Small Clone etc.
I remember the difference the ensemble effect used to make on the old Juno 106 we had at uni, it was literally like night and day! Everything just seemed so cold and one dimensional without it!
So, anybody got any pearls of wisdom??

Wow, odd. I guess it would be easier to add another BBD+ driver PCB and connect the specific points with cable.
Or make two CE-2's, connect and leave one LFO section out. But than you still need the inverting.

SAD1024's are good chips for what they are used for in the Electric mistress: going from jet flanging to mono chorus in a decent way.
Once you use two BBD lines, you should use cheaper BBD's AND get better results!

[SeanCostello]

I've got myself a nice Yamaha DX27 for christmas. Unfortunately it has only Mono outputs so i'm looking for a good chorus to thicken up the pads I want it for. In this case, a Dimension would be just what i'm looking for.

Try using the detune function of the DX27 to get some chorusing within the pads. In general, FM pads should use algorithms with as much parallelism as possible, with different amounts of detuning between the parallel operators. I can't find a listing of algorithm numbers, but algorithm #5 seems like a good start. You would want to detune the parallel carrier/modulator pairs, and then have internal detuning between the carrier the modulator for each pair, as this produces choral effects on its own. Avoid algorithm #8, as this seems to have no real FM, just parallel sine oscillators, which will be difficult to produce useful beating effects (well, you could have 2 beating sine waves, but that is pretty dull).

Sean Costello

[Jaicen_solo]

Whilst I can't argue with anything you've said there, none of it will help me to achieve a stereo output! In practical terms it's not really a big issue, but I want my pads to be lush and mellow

[Dave_B]

Whilst I can't argue with anything you've said there, none of it will help me to achieve a stereo output! In practical terms it's not really a big issue, but I want my pads to be lush and mellow

Bwanasonic had a good suggestion on page 2 of this thread.  He's used a Quadraverb, set to subtle pitch-shifting to achieve a nice stereo field.  Those things are pretty reasonable on eBay these days.  Maybe?   

[Jaicen_solo]

They are pretty reasonable, for the simple reason that after a few years, the internal battery fails. This results in a complete loss of all the internal programs, both factory and user presets. Effectively rendering them useless. That's why they're now flooding onto the market. Plus the fact that they're not actually that good
However, that is a good idea, apart from the above.